[Intel-gfx] [PATCH 6/7] drm/i915/display/adl_p: Correctly program MBUS DBOX A credits

[Souza, Jose]

On Fri, 2021-07-09 at 22:41 -0700, Matt Roper wrote:
> On Thu, Jul 08, 2021 at 02:18:26PM -0700, José Roberto de Souza wrote:
> > Alderlake-P have different values for MBUS DBOX A credits depending
> > if MBUS join is enabled or not.
> > 
> > BSpec: 50343
> > BSpec: 54369
> > Cc: Matt Atwood <matthew.s.atwood at intel.com>
> > Signed-off-by: José Roberto de Souza <jose.souza at intel.com>
> > ---
> >  drivers/gpu/drm/i915/display/intel_display.c | 16 ++++++++++++----
> >  1 file changed, 12 insertions(+), 4 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c
> > index 65ddb6ca16e67..fe380896eb99e 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display.c
> > +++ b/drivers/gpu/drm/i915/display/intel_display.c
> > @@ -3400,13 +3400,17 @@ static void glk_pipe_scaler_clock_gating_wa(struct drm_i915_private *dev_priv,
> >  	intel_de_write(dev_priv, CLKGATE_DIS_PSL(pipe), val);
> >  }
> >  
> > -static void icl_pipe_mbus_enable(struct intel_crtc *crtc)
> > +static void icl_pipe_mbus_enable(struct intel_crtc *crtc, bool joined_mbus)
> >  {
> >  	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
> >  	enum pipe pipe = crtc->pipe;
> >  	u32 val;
> >  
> > -	val = MBUS_DBOX_A_CREDIT(2);
> > +	/* Wa_22010947358:adl-p */
> > +	if (IS_ALDERLAKE_P(dev_priv))
> > +		val = joined_mbus ? MBUS_DBOX_A_CREDIT(6) : MBUS_DBOX_A_CREDIT(4);
> > +	else
> > +		val = MBUS_DBOX_A_CREDIT(2);
> 
> If we're in single-pipe / joined-mbus mode, then we'll program the
> credits to 6.  If we later turn on another pipe, reallocate the DDB, and
> turn off joined-mbus mode, we'll set that other pipe's credits to 4
> during the sequence of hsw_crtc_enable() -> icl_pipe_mbus_enable().  But
> don't we also need to go back re-program the credits down to 4 on the
> first pipe too (which is already enabled and won't be re-calling
> hsw_crtc_enable())?
> 
> I might be missing something; it's been a while since I really looked at
> any of the dbuf stuff...

skl_compute_ddb() is handling this cases, it will force a modeset in all pipes in cases like this.

> 
> 
> Matt
> 
> >  
> >  	if (DISPLAY_VER(dev_priv) >= 12) {
> >  		val |= MBUS_DBOX_BW_CREDIT(2);
> > @@ -3561,8 +3565,12 @@ static void hsw_crtc_enable(struct intel_atomic_state *state,
> >  	if (dev_priv->display.initial_watermarks)
> >  		dev_priv->display.initial_watermarks(state, crtc);
> >  
> > -	if (DISPLAY_VER(dev_priv) >= 11)
> > -		icl_pipe_mbus_enable(crtc);
> > +	if (DISPLAY_VER(dev_priv) >= 11) {
> > +		const struct intel_dbuf_state *dbuf_state =
> > +				intel_atomic_get_new_dbuf_state(state);
> > +
> > +		icl_pipe_mbus_enable(crtc, dbuf_state->joined_mbus);
> > +	}
> >  
> >  	if (new_crtc_state->bigjoiner_slave)
> >  		intel_crtc_vblank_on(new_crtc_state);
> > -- 
> > 2.32.0
> > 
> > _______________________________________________
> > Intel-gfx mailing list
> > Intel-gfx at lists.freedesktop.org
> > https://lists.freedesktop.org/mailman/listinfo/intel-gfx
>